1080
Photoionization of Anions in Rigid Media: Absorption and Emission Spectroscopy of Cyclooctatetraene Radical Anion, Its Ion Pairs, and Triple Ions Vladimh Dvoihk and Josef Michl*' Contribution from the Department of Chemistry, University of Utah, Salt Lake City, Utah 841 12. Received February 18. 1975
Abstract: Photoionization of salts of the cyclooctatetraene dianion (COT2-) with alkali metal cations (M+) in a 2-methyltetrahydrofuran glass at 77 K led to the formation of triple ions 2M+COT-, which were further transformed into ion pairs M'COT- and free COT- ions. Attempts to effect further ionization and to observe the so far hypothetical planar COT triplet were unsuccessful. Fluorescence from 2M+COT2-, 2K+COT-, 2Rb+COT-, and 2Cs+COT- was observed. In striking contrast, COT-, M+COT-, 2Na+COT- (and probably 2Li+COT-) do not emit. The absorption, emission, and ordinary as well as polarized excitation spectra indicate that COT- is planar and fairly rigid in the ground and lowest excited doublet state, at least in form of some of its ion clusters, and that the counterions are located on a fourfold or eightfold symmetry axis.
Photoejection of electrons from anions in a rigid glass is a well-known phenomenom2 The anions have been shown to be converted to the parent hydrocarbons (double anions to monoanions), the ejected electrons are trapped in the glassy solvent, and their presence can be detected by near-infrared absorption and by ESR spectroscopy. The nature and generality of the phenomenon a r e presently well enough understood that it appears entirely feasible to exploit it for preparation of novel hydrocarbons and radical ions from electronricher species. The latter a r e sometimes much more easily accessible, a classical example being the pentalene dianion which has been known for more than a decade, while pentalene radical anion is unknown, and pentalene itself appears to be extremely e l ~ s i v e . ~ This approach to new hydrocarbons makes sense particularly in instances in which their expected reactivity is so high that conventional techniques fail and matrix-isolation studies followed by spectral techniques a r e called for. In such cases it is very useful to have several precursors available and a n electron-richer ion might well be one of these. Reactivity studies may be possible after the glass is melted if the ejected electrons a r e first scavenged by a suitable trapping agent. It is also worth noting that, a t least in principle, in photoejection of an electron from a doublet ground state of a radical anion with light of suitable wavelength, spin selection rules permit not only the formation of a singlet but also of a triplet state of the neutral product. The possibility of opening a new pathway to triplet molecules provides an additional incentive for the present line of investigation. Since some anions and many double anions will be present in the form of tight ion pairs or triple ions in glassforming solvents such as 2-methyltetrahydrof~ran,~.~ photoejection will usually not lead to perturbation-free hydrocarbons or radical anions. This may be considered a disadvantage if preparation of the free species is the goal, but on the other hand, the availability of a choice of counterions permits an observation of the new species under a variety of minor perturbations and this can be very valuable, particularly in electronic spectroscopy, where a study of perturbation-induced shifts should often allow a differentiation of spectral features due to vibrational sublevels from those due to several independent electronic transitions. The effects of nearby positive charges on properties of neutral hydrocarbons might be of considerable interest in themselves in view of their possible relation to adsorption phenomena. Also, clusters of two positive ions with an anion carrying a single negative charge a r e not easily accessible otherwise. Journal of the American Chemical Society
/
98.5
/
Before enga ing in a n investigation of novel hydrocarbons and ions,3' we have decided to reaffirm the generality and applicability of the method on a model system in which the product was known but its electronic spectrum had not been studied extensively. W e have chosen the cyclooctatetraene dianion (COT2-), which is expected to give the radical anion (COT-) in the first step. The uv absorption spectrum of COT- in liquid ammoniaS and its E S R spectrum5-' have been published. While our work was in progress, the absorption curve in 2-methyltetrahydrofuran (2-MTHF) was also published,* as was a report that photoejection from COT2- can be accomplished in fluid solutions and monitored by E S R spectroscopy.6 T h e choice of COT2- for our first experiments was dictated by two additional considerations. First, it is important to find out how firmly bound the electrons can be before the photoejection fails to occur with photons for which glassy 2 - M T H F is still transparent (A > ca. 235 nm), since this ought to be perhaps the most serious limiting factor for the present version of the method. This photoejection threshold should be related to the energy of the highest occupied molecular orbital ( H O M O ) and should depend on the presence of counterions. In COT2-, the H O M O is nonbonding, whereas it is antibonding in most of the hydrocarbon ions studied so far, so that COT2- is a fairly difficult test case. Second, if it turned out that both electrons can be photoejected, C O T itself should be formed. It seemed to us that there was a slight chance that COT might be formed in a vertical process as a planar octagonal triplet which, at that geometry, could represent the ground state and thus be metastable with respect to conversion to the ordinary nonplanar singlet. Such a species would be of considerable theoretical interest. The present paper reports a preparation of COT- in rigid 2 - M T H F by photoejection from COT2- and describes its absorption and ordinary and polarized fluorescence emission and excitation spectra in the presence of one or two alkali metal counterions. Attempts to prove the presence of planar triplet COT have given only ambiguous results.
Experimental Section Solutions of COT2- salts in 2-MTHF were obtained by reaction of COT in 2-MTHF with metal mirrors or metal chunks (Li) using standard high-vacuum techniques. 2-MTHF (MCB chromatoquality reagent) was refluxed with Na-K alloy (MSA Research Corp.) for several hours, fractionally distilled through a 3-ft column, transferred into a vessel containing LiAIH4 or Na-K alloy, degassed, and distilled in vacuo into a storage bulb containing tetra-
March 3, 1976
1081
A(nm)
Figure 1. Absorption spectra in 2-MTHF at 77 K (3-mm path length): (A) baseline for spectra B and C below 400 nm; (B) (1.5 f 0.3) X 4 2K+COT2-; (C) 2K+COT- e-(solv), obtained from B by 254-nm irradiation (90 i 10% conversion). The right-hand side of curve C (above 4 0 nm) has been corrected for baseline. The insert between 1000 and 1500 nm has been shifted down by 1.0 unit of O.D. (D) A more concentrated soM 2K+COT2-. lution of 2K+COT- e-(solv), obtained by irradiation of ca. 5 X
+
+
cene and Na-K alloy. COT (BASF) was freshly vacuum-distilled just before use. Even minute traces of moisture apparently lead to formation of some 1,3,5-cyclooctatrienewhen C O T in 2-MTHF is reduced over an alkali metal mirror. This is not immediately obvious in the uv absorption spectrum, but after irradiation 1,3,5,7-0ctatetraene is formed and interferes badly. The simplest way in which this problem can be avoided is to keep the COT-2-MTHF solution for a day or two at -80' at the bottom of a sealed vessel containing a metal mirror on the wall above the solution. During this time, a part of the mirror is destroyed, presumably by water vapor. Subsequent contact of the COT-2-MTHF solution with the mirror gives a COT2- solution without detectable traces of impurities. The COT2- solutions thus prepared were about 2.10-2 M in COT2and were then diluted by standard techniques for spectroscopic measurements. Break-seals were used for low-temperature addition of further reagents where called for (crown ether). Authentic samples of 1,3,5,7-octatetraene were prepared by photolysis of 1,3,5-cyclooctatriene~which in turn was prepared from COT.'O Absorption spectra were taken with a Cary 17 spectrophotometer. Low-temperature measurements were performed in a Suprasil cell immersed in a Dewar vessel with Suprasil windows filled with liquid nitrogen. Baseline corrections were performed routinely and were important especially in the near-ir region where strong harmonics of solvent ir bands occur. The emission measurements were done using a I-kW Xe arc with a regulated power supply, a 100-mm water filter, 250." focal length monochromators (Schoeffel Instruments Co.), appropriate focusing mirrors, a Centronics Q4283R photomultiplier with S-20 response, a Keithley Instruments Model 224 power supply, a PAR Model 184 photometric preamplifier, a PAR Model 124 lock-in amplifier, and Omnigraphic Model 2000 Houston Instruments X-Y recorder. Exciting light was chopped at 13 Hz using a PAR Model 125 chopper whose output was used as an external reference for the lock-in amplifier. Front-surface excitation and 25O angle between exciting and emitted beams were used, except in polarized emission measurements, in which the angle was 90'. I n the latter measurements, a polarization scrambler was used before the first polarizer to equalize the exciting light intensities in both polarizations. Polacoat polarizing sheets were used as polarizers. ESR measurements were performed on a standard V-line Varian X-band spectrometer with 100-kHz modulation using a 9-in.
magnet equipped with Fieldial control. The Klystron frequency was about 9.5 G H z for room-temperature measurements and about 9.15 G H z for low-temperature measurements. The latter were performed in a Dewar insert of a Varian variable-temperature accessory. The ESR cavity was a standard universal Varian 104 mode cavity equipped with irradiation slots. For irradiation of samples, a 200-W high-pressure Hg lamp (PEK) was used in conjunction with a 100-mm water filter, interference filters (narrow band transmission, Balzers, and wide-band reflection, Fish-Shurman). and cut-off filters (Schott). For 254mm irradiation, a low-pressure Hg lamp (Ultraviolet Products, Inc.) was used without any filters. For shorter wavelength irradiation, Philips metal vapor lamps were used without any filters (Zn, 214 nm; Cd, 229 nm). In order to minimize the effects of destruction of COT- by capture of solvent-trapped electrons liberated by excitation with visible light emitted by these lamps, the sample was simultaneously irradiated by a low-pressure Hg lamp.
Results COT2-. Excitation and Emission Spectra. The uv absorption spectrum of COT2- has been known for a long time." Even at 77 K (2-MTHF), it is very diffuse. It starts with a weak shoulder near 390 nm and increases to a stronger shoulder near 285 nm (Figure 1). Their exact positions depend slightly on the counterion. We now wish to report observation of emission from rigid solutions of COT2- in 2M T H F a t 77 K. It is broad and structureless, starts near 400 nm, and reaches a broad maximum near 530 nm. It is unlikely that the emission originates in a species other than COT2-, not only because of its wavelength, but also since it does not change when the excitation wavelength is changed (310-400 nm), the position of the maximum depends only slightly on the nature of the counterion ( N a + , K+, Cs+), and the excitation spectrum (240-400 nm) follows the absorption spectrum of COT2- at long wavelengths. However, it shows a sudden drop (15-25%) near 335 nm, essentially independently of the nature of the counterion. At shorter wavelengths, it again continues normally, albeit with reduced intensity. The emission is not observed when C O T itself is irradiated in 2-MTHF at 7 7 K. Since the absorption and excitation spectral curves in-
DuoFbk. Michl
/
CyclooctatetraeneRadical Anion
1082 crease rather rapidly in the vicinity of 340 nm, the sharp drop near 335 nm gives rise to an apparent sharp peak a t 340 nm. We have checked carefully that this peak is indeed a n inherent feature of the excitation spectrum of COT2(the spectrum is independent of the monitoring wavelength, and emission obtained by excitation a t 340 nm is identical with that obtained a t other wavelengths). This sudden decrease in the quantum yield of fluorescence should correspond to the onset of a competing physical or chemical process, and we assign this tentatively as the photoionization process COT2COTe- in agreement with the result discussed in more detail below. While irradiation with longer wavelengths is without effect under our experimental conditions, light of X E 340 nm or shorter causes electron ejection. The quantum yield of photoionization clearly is much less than one even a t quite short wavelengths (250 nm) since, even a t these excitation wavelengths, emission is clearly seen. Extended irradiation with 254-nm light destroys all COT2- and replaces it by COTand trapped electrons, as shown by the absorption spectrum and discussed in detail below; a t the same time, the green emission attributed here to COT2- disappears, providing further support for our assignment. It is unlikely that the emission originates in recombination of photoelectrons with electron traps since quite different emissions were observed with other dianions in 2-MTHF glass. It should be noted that our absorption, emission, and ionization results for COT2- refer to the species predominant in a 2 - M T H F solution a t 77 K. As is discussed in more detail below, this is the double ion pair 2M+COT2-, with the probable exception of the lithium salt in whose solution several species a r e apparently present simultaneously in significant concentrations. COT-. Preparation. Irradiation of dilute (1 0-3- 10-5 M ) glassy solutions of alkali metal salts of COT2- in 2 - M T H F a t 77 K with light of wavelengths approximately 340 nm or shorter (interference filter, 200-W H g arc) causes disappearance of the uv absorption spectrum of COT2- and appearance of the typical broad absorption band of solvated electrons with two maxima near 1200 and 1380 nm and of bands very similar to those reported5s7 for COT- in the 300-400-nm region (Figure 1). A high degree of conversion can be achieved, quite close to 100%judging by the disappearance of the absorption peak of COT2- a t 285 nm, where the COT spectrum has a minimum. An isosbestic point is observed. In most of our experiments, 254-nm light was used for the photoionization. This wavelength is absorbed strongly by COT2- and only weakly by COT-. In the E S R spectrum, the typical easily saturated broad signal of trapped electrons is observed, as well as the seven broad lines of COT- a t 3.143 separation (the outer two lines of the nonet a r e weak and difficult to observe). An identical E S R spectrum was observed on a frozen sample of a solution containing chemically prepared COT- (see below) and exhibiting a sharp nonet a t room temperature (3.1 5-G splitting). N o apparent changes occur upon standing in the dark a t 77 K. Irradiation with visible or near-infrared light, well known to liberate electrons from solvent traps,I2 reduces the bands of the new species in the uv and those of electrons in near-ir and ESR spectra and eventually almost completely restores the spectrum of COT2-. Reconversion to COT2can be also achieved by melting the glass. The cycle is almost fully reversible and can be repeated many times. In order to achieve complete conversions to COT-, it is important to use narrow-band uv radiation since visible light liberates electrons from their traps and establishes a photostationary state. After extended periods of irradiation, the total amount of available electrons decreases, presum-
-
Journal of the American Chemical Society
+
/
98:5
ably due to reaction with solvent, and this apparently causes the observed slight degree of irreversibility in the cyclic process. Trapping of photoejected electrons by dimethyl ether and other ethers under similar conditions has been reported.I2 Already very small traces of impurities change the behavior of the sample considerably. Uv irradiation of old samples, imperfectly dried samples, and samples prepared with very reactive metals (Rb, Cs) gives mixtures containing 1,3,5,7-0ctatetraene and a t least one red-colored species in addition to COT- and e-(solv). The octatetraene was identified by comparison of its uv, fluorescence, and fluorescence excitation spectra with a n authentic sample. We believe that it is formed photochemically from l ,3,5-cyclooctatriene in a well-known r e a ~ t i o nPresumably, .~ in these impure samples some cyclooctatriene is formed by protonation of COT2- and its broad absorption band a t 265 nm, unlike the sharp peaks of 1,3,5,7-0ctatetraene, cannot be observed since it is hidden under COT2- absorption. The bands in the visible region due to the red-colored species increase considerably upon removal of electrons from their traps by red light (700 nm). Upon subsequent irradiation with blue light (400 nm), the bands of the red-colored species as well as COT- almost disappear, but a band of solvated electrons does not appear. W e believe that the red species are radical anions of the olefins present and that red photons do not have sufficient energy to photoionize them while the blue ones do, and that the ejected electrons are trapped by COT- present in large excess. Firm identification of the red species will require further work, COT-. Ion Pairing. While the overall appearance of the COT- spectrum depends little on the counterion used, closer inspection reveals significant differences in peak intensities and positions. The intensity of the sharp bands a t 320400 nm relative to the broad peak a t 310 nm is high for Cs+, Rb+, and K+ counterions (Figure l ) , lower for N a + and Li+, undoubtedly, a t least partly because the former have smaller bandwidths. The position of the band origin shifts a few nanometers to the red each time the radius of the counterion increases going down the periodic table (Na’, 383.5; K+, 287; Rb+, 389; Cs+, 392 nm), but the fine-structure spacing remains almost unchanged (Figure 2), confirming that all of the peaks between 320 and 400 nm belong to the same electronic transition, as had been suspected on the basis of calculation^.^ According to N M R studies, the Cs+, Rb+, K+, and N a + salts of COT2- exist as doubly contact ion pairs in 2M T H F a t room temperature.13 They probably also are in this form a t 77 K since only one form of the photoproduct is formed initially in our experiments, assigned as 2M+COT-. However, prolonged irradiation with uv light or very slight softening of the glass reduces its concentration, and a new species appears with identical vibrational structure but an origin shifted slightly to the red. Only one such new form is observed distinctly with Cs+ and Rb+, but two of them have been observed with N a + and K+ (Figure 2). W e propose that the new species are formed by successive diffusion of the counterions away from COT-. Such a process has been observed before on other hydrocarbon anions upon glass softening.2b Similarly, we attribute the effect of extended uv irradiation to “internal melting”-conversion of electronic energy of the absorbed quanta into vibrational energy causing local heating and decrease in viscosity. The first peak of the longest wavelength species is a t 396 nm irrespectively of the original counterion, and we assign it as the free COT- anion in 2-MTHF solvent. The peak positions are in perfect agreement with those published recently by Shida and Iwata8 for the free COT- ion in 2-MTHF. The other shifted species is then assigned to the tight ion pair
/ March 3, 1976
1083
-
I
ABS
EXP
+
EXC
EM
ASSIGNMENT
COT
i'
~___
LI
-
I
1
I
\ I
MT- L,'
COT
-
COT- 2 N i * COT- NO' COT-
m t-7
____ ,r _ _,,--I
4
5c
COT'ZK' COT-K' COT -
K
I
4c
COT- 2Rb'
'
Rb
I
cs
--+mhl-l'
